[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

RE: layer interactions (was UTF-8+names)

• To: xml-dev@l...
• Subject: RE: layer interactions (was UTF-8+names)
• From: "Simon St.Laurent" <simonstl@s...>
• Date: Wed, 22 Oct 2003 10:46:11 -0400
• In-reply-to: <007701c39811$634f4030$6401a8c0@pcukmka>

michael.h.kay@n... (Michael Kay) writes:
>> I've done some experiments with this in Ripper, and it's 
>> about ten lines of code to make the change.  
>
>Presumably you create a dependency on attribute order by requiring a
>namespace to be declared before it is used in an entity reference?
>Otherwise you open the door to all sorts of circularities.

I process start tags for namespace declarations before reporting their
contents, so that shouldn't be a problem.

Well, except for one perverse case: a namespace declaration which uses
an entity in its value, an entity which uses the namespace that
declaration is declaring.  That would look like:

xmlns:e="http://simonstl.com/ns/&e:hahaha;"

Short of that - which I think is easily banned - no, it's not a problem.
Attribute order doesn't raise an issue any more than the use of
namespaces on attributes raises a problem when the namespace declaration
appears after the attribute.

Namespaces have plenty of perversity, but once you've been through a
round of dealing with that perversity, it's not hard to proceed from
there.

• References:
  • RE: layer interactions (was UTF-8+names)
    • From: "Michael Kay" <michael.h.kay@n...>

• Prev by Date: Re: UTF-8+names
• Next by Date: Re: Alternative "character entity" proposal
• Previous by thread: Re: layer interactions (was UTF-8+names)
• Next by thread: Re: UTF-8+names
• Index(es):
  • Date
  • Thread